Recombination, Solvation and Reaction of CN Radicals Following

Article pubs.acs.org/JPCA

Recombination, Solvation and Reaction of CN Radicals Following Ultraviolet Photolysis of ICN in Organic Solvents Philip Coulter,† Michael P. Grubb,† Daisuke Koyama,† Igor V. Sazanovich,‡ Gregory M. Greetham,‡ and Andrew J. Orr-Ewing*,† †

School of Chemistry, University of Bristol, Cantock’s Close, Bristol BS8 1TS, U.K. Central Laser Facility, Research Complex at Harwell, Science and Technology Facilities Council, Rutherford Appleton Laboratory, Harwell Oxford, Didcot, Oxfordshire, OX11 0QX, U.K.

‡

S Supporting Information *

ABSTRACT: The fates of CN radicals produced by ultraviolet (UV) photolysis of ICN in various organic solvents have been examined by transient electronic and vibrational absorption spectroscopy (TEAS and TVAS). Near-UV and visible bands in the TEAS measurement enable direct observation of the CN radicals and their complexes with the solvent molecules. Complementary TVAS measurements probe the products of CN−radical reactions. Geminate recombination to form ICN and INC is a minor pathway on the 150 fs −1300 ps time scales of our experiments in the chosen organic solvents; nonetheless, large infrared transition dipole moments permit direct observation of INC that is vibrationally excited in the CN stretching mode. The time constants for INC vibrational cooling range from 30 ps in tetrahydrofuran (THF) to 1400 ps in more weakly interacting solvents such as chloroform. The major channel for CN removal in the organic solvents is reaction with solvent molecules, as revealed by depletion of solvent absorption bands and growth of product bands in the TVA spectra. HCN is a reaction product of hydrogen atom abstraction in most of the photoexcited solutions, and forms with vibrational excitation in both the C−H and CN stretching modes. The vibrational cooling rate of the CN stretch in HCN depends on the solvent, and follows the same trend as the cooling rate of the CN stretch in INC. However, in acetonitrile solution an additional reaction pathway produces C3H3N2• radicals, which release HCN on a much longer time scale. dynamics36 has precipitated numerous experimental studies of the chemical dynamics of ICN photolysis in liquids using ultrafast transient absorption spectroscopy.37−42 The interpretation of these experimental measurements has been supported by simulations of the dissociation dynamics and fates of the I and CN fragments.41,43−49 Rivera et al.39 reported a peaked feature centered at 390 nm in the UV/visible absorption spectrum after photolysis of ICN in water and ethanol that closely resembled a low resolution gas-phase B ← X spectrum of the CN radical. This peak was assigned to free, unsolvated CN radicals, and rapidly evolved into significantly broader spectral features peaking at 326 nm in water and 415 nm in ethanol which correspond to solvated radical species. A combination of these experimental measurements and dynamics simulations using the potential energy surfaces of Morokuma and co-workers4,6 showed that CN and I radicals will undergo recombination as a result of the solvent cage in ∼60 fs and diffusive recombination in ∼16 ps.39 Furthermore,

1. INTRODUCTION The ultraviolet photolysis of ICN is an extensively studied process which has been used to explore nuclear dynamics in electronically excited states and bimolecular reactions of the photofragment CN radical. The dissociation of the I−CN bond in isolated, gas-phase ICN molecules has been directly observed on the femtosecond time scale,1−3 and computational and experimental studies characterized the dissociative electronic potentials4−6 and the resulting quantum-state specific I and CN energy distributions.7−35 The ultraviolet absorption of gaseous ICN peaks at wavelengths around 250 nm, with the broad absorption band structure composed of excitations from the linear 1Σ+0 ground state to excited states of 1Π1, 3Π0+ and 3Π1 symmetry. The 3Π0+ state correlates asymptotically with CN (X2Σ+) radicals and spin−orbit excited I*(2P1/2) atoms, but a conical intersection with the 1Π1 state provides a nonadiabatic pathway to the lower-lying CN (X2Σ+) + I(2P3/2) asymptote.4 This conical intersection is accessed within ∼30 fs, and the dynamics of passage through the conical intersection induce high degrees of rotational excitation in the CN radical. The suggestion by Benjamin and Wilson that ICN would provide a model system for the study of solution phase © XXXX American Chemical Society

Received: November 2, 2015 Revised: November 27, 2015

A

DOI: 10.1021/acs.jpca.5b10716 J. Phys. Chem. A XXXX, XXX, XXX−XXX

Article

The Journal of Physical Chemistry A

were selected in a random order and were controlled with subpicosecond time resolution and maximum values of up to 3000 ps (ULTRA) and 1300 ps (Bristol). The IR laser pulse wavelength was centered near 2100 cm−1, and had an approximate bandwidth of 500 cm−1 (ULTRA) or 300 cm−1 (Bristol). IR radiation transmitted through the sample was dispersed onto a 128-element array detector (with mercury cadmium telluride elements) from which averaged spectra were accumulated and processed to determine the wavelengthdependent changes in absorbance induced by the UV pump pulse. Spectra of 1,4-dioxane were used to calibrate the wavelength scale of the TVAS measurements. Subsequent analysis of TVAS data used the KOALA software package.61 Transient electronic absorption spectroscopy (TEAS) measurements were made using the Bristol ultrafast laser system. In addition to the samples used for TVAS, the TEAS measurements used 0.29 M solutions of ICN in methanol (Fischer Chemicals, 99.99%) and ethanol (BDH Reagents, Spectroscopy grade). Approximately 1 μJ/pulse of the fundamental output of an amplified Titanium:Sapphire laser (1 kHz, 40 fs pulse duration) generated white-light continuum (WLC) pulses by focusing into a rastered 5 mm thick CaF2 window using a 100 mm focal length lens. The recollimated WLC pulses spanned the approximate wavelength range of 320 to 650 nm. Time delays between the UV pump and WLC probe pulses were selected using the same procedure as for TVAS measurements. The near-UV and visible radiation components of the WLC pulses transmitted by the sample were collected into an optical fiber and dispersed onto a 750pixel CCD spectrometer (Avantes, AvaSpec-DUAL) with a spectral resolution of 0.6 nm. Analysis with the KOALA package61 corrected spectra for probe−pulse frequency chirp and enabled decomposition into spectral components. Steady state FTIR spectra of samples before and after UV irradiation identified the positions of the CN stretching bands of ICN and HCN dissolved in the various solvents. The electronic absorption bands of ICN in different solvents were characterized by subtracting steady state UV/vis spectra of the solvent from the corresponding spectra of the ICN solutions.

Bradforth and co-workers demonstrated that nascent, rotationally hot CN radicals will retain their rotational coherence for up to a few picoseconds in solution.38,49 The photochemistry of ICN has also been studied in cryogenic matrices, with Fraenkel and Haas assigning an infrared band that developed to lower wavenumber than the ICN fundamental CN stretching mode to the INC isomer.50 In addition to recombination, CN radicals produced through photolysis in solution may undergo bimolecular reactions.51−53 Hochstrasser and co-workers were the first to apply transient infrared absorption spectroscopy to observe the production of HCN and DCN in chloroform and chloroform-d on ultrafast time scales.54 Further studies by Crowther et al.55,56 identified the role played by CN radical complexes with chlorinated solvents, while Orr-Ewing and co-workers observed modespecific vibrational excitation of the C−H stretching and bending modes of HCN products of CN−radical reactions, and their subsequent cooling by coupling to the solvent bath.41,42,48,57,58 The current study combines ultrafast transient absorption spectroscopy in the UV/visible and infrared spectral regions to observe the initial solvation of CN radicals from ICN photolysis, and the competition between geminate recombination to form ICN and the isomeric INC, solvent−complex formation, H atom abstraction reactions, and addition reactions with solvent molecules. We report observations following ICN photolysis in chloroform, dichloromethane (DCM), acetonitrile, tetrahydrofuran (THF), methanol, and ethanol, providing a comprehensive picture of the various fates of photoproduced CN radicals in solution on femtosecond to picosecond time scales.

2. EXPERIMENTAL DETAILS Ultrafast transient absorption spectroscopy measurements used both the ULTRA laser system, at the Central Laser Facility of the Rutherford Appleton Laboratory, and a femtosecond laser system, at the University of Bristol. Detailed descriptions of the laser-based instruments and the sample preparation and handling procedures are presented elsewhere,58−60 and we provide only a brief summary here. Transient vibrational absorption spectroscopy (TVAS) experiments were conducted with 0.29 M solutions of ICN (98%; Acros Organics) and BrCN (Aldrich; 97%) which were prepared in acetonitrile (Fisher Chemical; 99%), dichloromethane (VWR Chemicals; > 99%, with 0.2% ethanol), chloroform (VWR-Chemicals, > 99%), tetrahydrofuran (VWR, 99.5%), and d-acetonitrile (Sigma-Aldrich; 99.8% D). All hydrogenated solvents were dried using 3 and 5 Å zeolite molecular sieves. ICN was recrystallized from toluene (SigmaAldrich Analytical grade) before use. Sample volumes of 5−10 mL were circulated through a Harrick cell using a peristaltic pump and PTFE tubing. The Harrick cell was fitted with CaF2 windows separated by 380 μm. ICN was photolyzed using a UV pump pulse (approximate duration: 50 fs ULTRA; 120 fs Bristol) with a wavelength of 267 nm. The pump pulses were delivered to the liquid samples at repetition rates reduced to half the laser output frequency (output frequency: 10 kHz ULTRA; 1 kHz Bristol) using a mechanical chopper. For TVAS measurements, IR pulses at the laser output frequency were overlapped spatially with the UV pulses at the sample. A motorized translation stage varied the length of the path followed by the UV laser pulses to set the time delays between UV pump and IR probe pulses. Delays

3. RESULTS AND DISCUSSION The UV absorption bands of ICN in various solvents were measured by steady state absorption spectroscopy on static samples to identify any solvent-induced shifts and changes to the band profiles. These results are presented first, and compared to the known gas-phase spectrum of ICN.62 TEAS was then used to observe the production, solvation and loss of CN radicals following ICN photolysis in solution. The fates of these radicals were studied using TVAS in the mid-IR spectral region. 3.1. Steady-State UV Absorption Spectra. The UV absorption bands of ICN are well-characterized in the gas phase but are modified by interaction with solvent molecules. Figure 1 compares the gas-phase spectrum63 with spectra obtained in various solvents, for which the solvent-only absorption spectra have been subtracted to isolate the ICN features. The maxima in the solution spectra are shifted to lower wavelength than the gas-phase spectrum, with the shift increasing from chloroform and dichloromethane to acetonitrile, methanol and THF. These spectral shifts have previously been attributed to stabilization of the ground state of ICN, which has a large permanent dipole moment, by polar solvents. To a lesser extent, these solvents also destabilize the electronically excited states of ICN B

DOI: 10.1021/acs.jpca.5b10716 J. Phys. Chem. A XXXX, XXX, XXX−XXX

Article

The Journal of Physical Chemistry A

In each case, absorption at 267 nm is on the long-wavelength side of the ICN band and favors excitation to the 3Π0+ and 3Π1 states correlating adiabatically to the I* + CN and I + CN dissociation limits, respectively. Although there is a shallow minimum in the 3Π0+ state (and perhaps also in the 3Π1 state), dissociation is prompt (∼100 fs) at the excitation energy used.39 3.2. TEAS of Photoexcited ICN Solutions. The timeresolved electronic absorption spectra of ICN solutions show transient absorption features across the near-UV and visible regions, and examples are presented in Figure 2 for five solvents: chloroform (Figure 2a), dichloromethane (Figure 2c), acetonitrile (Figure 2d), ethanol (Figure 2e) and methanol (Figure 2f). The interpretation of the broad overlapping bands was previously discussed by both Bradforth and co-workers39 and Keiding and co-workers.40 Despite this prior analysis, the decomposition of such spectra and the subsequent assignment of component bands to specific molecular species remains a challenge. Nonetheless, ICN photolysis in chloroform provides a clear separation of some of the observed absorption features; the decomposition of one transient spectrum obtained at a time delay of 1.2 ps is presented in Figure 2(b). The sharp peak centered at 390 nm is assigned to CN radicals which are only weakly solvated by chloroform, resulting in a small shift of the (B ← X) transition from the gas phase absorption center. We refer to these weakly solvated CN radicals as f ree CN and consider their electronic states to be largely unperturbed by interactions with the solvent. The free CN peak grows rapidly

Figure 1. Normalized steady-state UV absorption spectra of ICN in various common solvents. Pure solvent spectra have been subtracted to isolate the contributions from ICN absorption. The solvents are as follows: THF (orange); methanol (purple); acetonitrile (red); chloroform (green); dichloromethane (blue). A gas-phase spectrum is shown by dots, with absorption cross sections on the right-hand axis.

contributing to the broad UV absorption band.38,39,47 However, the band maxima do not show a straightforward correlation with either solvent molecular dipole moments or relative permittivities, suggesting a more complex interaction of the solvents with the electronic states of ICN.

Figure 2. TEAS spectra of ICN photolysis in (a) chloroform, (c) dichloromethane, (d) acetonitrile (e) ethanol, and (f) methanol. The various colors from purple to dark red represent spectra obtained at increasing excitation-to-probe laser pulse time delays up to 1300 ps, as indicated by inset keys. Panel b provides an example of the decomposition of a transient absorption spectrum obtained in chloroform at a single time delay of 1.2 ps. In the decomposition, the feature peaking at 340 nm (red line) has contributions from both solvent-to-solute I(2P3/2) CT and CN−radical complex absorption bands. The shape of this feature is taken from a late time spectrum for ICN photolysis in chloroform. The feature centered at 390 nm (blue line) is a Lorentzian function that represents free CN−radical absorption, and the broad feature centered at 540 nm is a Gaussian function representing solvent-to-solute I*(2P1/2) CT. Similar contributions to the TEA spectra can be identified for the measurements made in acetonitrile and dichloromethane, but are less distinct in ethanol and methanol where the broad feature representing I(2P3/2) CT and CN−radical complex absorption peaks between 375 and 395 nm. C

DOI: 10.1021/acs.jpca.5b10716 J. Phys. Chem. A XXXX, XXX, XXX−XXX

Article

The Journal of Physical Chemistry A before decaying with a 3.3 ± 1.0 ps time constant; meanwhile, a broad feature peaking at 340 nm grows with a 1.7 ± 0.7 ps time constant before decaying slowly across the experimental time window. One plausible assignment of this feature is to a solvent-to-solute charge transfer (CT) band in which the solute is a ground-state iodine atom (I(2P3/2)). The corresponding I*(2P1/2)−solvent CT band is shifted to longer wavelength and contributes a broad background to the transient absorption spectra that peaks in the wavelength region above 450 nm. Electronic quenching of the spin−orbit excited I* atoms will reduce the intensity of this broad feature and should cause a commensurate rise in the I(2P3/2)−solvent CT band. The time constants for the growth of the 340 nm features are similar in chloroform (1.7 ± 0.7 ps) and dichloromethane (1.1 ± 0.5 ps), but are slightly smaller than the time constants of 2.0 ± 0.4 ps and 1.4 ± 0.5 ps for the initial fast decay of the I*(2P1/2) CT band. Bradforth and co-workers39 proposed that following ICN photolysis in water, a broad solvated-CN absorption feature overlaps the I(2P3/2) CT band in the near UV, with the two spectral features being difficult to distinguish. Much of the absorption observed in our TEA spectra at near-UV wavelengths and extending to the visible may therefore be associated with solvated forms of the CN radical. The shift of the CN absorption to shorter wavelengths as a result of solvation has been quantified theoretically by Pieniazek et al.64 The smaller blue shift of the CN absorption in ethanol than in water is attributed to ethanol’s lower dipole moment. The broad nearUV feature cannot be assigned uniquely to an I(2P3/2)−solvent CT band or solvated CN on the basis of our spectroscopic observations; instead, both species are likely to contribute to the absorption at 340 nm, and hence to the time constant for its growth. The low dipole moments of chlorinated solvents and acetonitrile compared to water do not obviously account for the spectral position of the solvated CN bands to the blue of the free CN (B ← X) absorption; instead, strong complexing interactions of CN-radicals with individual solvent molecules are thought to be responsible, as proposed by Crowther et al.55,56 We explored the spectra of CN−solvent complexes in chloroform, dichloromethane and acetonitrile further by computational methods. Density functional theory (DFT) calculations were used to optimize the structures of the complexes, and time-dependent DFT (TD-DFT) calculations predicted CT absorptions peaking at 336 nm, 330 and 294 nm for CN−chloroform, CN−DCM and CN−acetonitrile complexes (see the Supporting Information). These calculations were performed using the Gaussian 09 package65 with the unrestricted CAM-B3LYP functional and an aug-cc-pVTZ basis set. The calculations are in reasonable agreement with experimental observations for CN−chloroform and CN− dichloromethane TEA spectra, and the prediction of a shorter absorption wavelength for the CN−acetonitrile complex accords with the spectra shown in Figure 2d. The peak of the CN−acetonitrile absorption lies further into the UV than the extent of our white-light continuum, but the long-wavelength wing of this band is observed. Further resolution of the chemical identities of the species responsible for the 340 nm band comes from comparative studies of BrCN and ICN photolysis in acetonitrile solution. The same broad near-UV absorption band is observed when ICN is preplaced by BrCN, indicating an assignment to solvated CN radicals, but Br-solvent CT bands cannot be

Figure 3. Normalized intensity of the broad feature peaking at 340 nm plotted as a function of time (dots) after (a) BrCN and (b) ICN photolysis in acetonitrile solution. The intensities were fitted with a single exponential decay for BrCN (blue line) and a double exponential decay for ICN (red line).

immediately discounted. Figure 3 compares the time dependence of the intensity of the 340 nm feature after BrCN (Figure 3a) and ICN (Figure 3b) photolysis in acetonitrile. The 340 nm feature decays to 20% of the band maximum after BrCN photolysis, with a single exponential time constant of 11 ± 2 ps which is similar to the 9.8 ± 1.5 ps initial time constant measured after ICN photolysis. The longer component of the decay evident after ICN photolysis (with time constant of 158 ± 25 ps) is not observed after BrCN photolysis, and is therefore assigned to the I(2P3/2)−solvent CT band. The longer-lived, but weak absorption observed for the BrCN solution is suggested to be a signature of the corresponding CT band for Br−acetonitrile complexes. The faster decay seen in both measurements most likely corresponds to loss of solvated CN-complexes, which undergo reaction or geminate recombination with an ∼10 ps time constant. TEAS measurements subsequent to ICN photolysis in dichloromethane show similar spectral features to those observed in chloroform. However, the band assigned to free CN is only visible at the earliest time delays in acetonitrile, ethanol and methanol; for these solvents, the spectra are dominated by broad absorption bands. The photochemistry of ICN in acetone was discussed recently,66 and shows many of the features described above, but an additional band centered below 350 nm was observed that was assigned to absorption from the lowest triplet state of acetone. The 340 nm band does not return to the baseline within the 1300 ns time scale of the experimental measurements in any of the organic solvents studied, but does disappear in water where D

DOI: 10.1021/acs.jpca.5b10716 J. Phys. Chem. A XXXX, XXX, XXX−XXX

Article

The Journal of Physical Chemistry A fast radical recombination dominates.39 Although the reaction of CN radicals with solvent molecules to make HCN may be relatively slow in chlorinated solvents,55,56 in most other organic solvents the CN radicals are expected to react on picoto nanosecond time scales, producing an organic radical coproduct. The corresponding reaction is endothermic by ∼24.8 kJ mol−1 with water67 and is therefore unfavorable. Spin−orbit excited I*(2P1/2) atoms will quench rapidly to the ground state I(2P3/2) atoms, and these in turn decline steadily in numbers because of diffusive recombination with other I atoms or organic radicals. These I atoms are likely contributors to the late-time 340 nm absorption band as argued above. However, possible overlap of the broad 340 nm band with other spectral features, and uncertainty over the expected shapes and center wavelengths of the I-solvent and I*-solvent CT bands precluded further interpretation. Table 1 summarizes the time constants for growth and decay of species contributing to the TEAS data in the solvents

spectroscopy of steady state solutions, and has a negative intensity (a “bleach” feature) in the transient spectra because of depletion of ICN by the UV photolysis laser. Similar bleaches appear following ICN photoexcitation in each of the solvents considered. These peaks show limited signs of recovery on the time scales of our measurements, indicating that liberated CN mostly avoids geminate recombination with I atoms and is instead removed by processes such as reaction with the solvent molecules. In DCM, the depth of the ICN bleach feature grows with increasing time delay, which might be construed as evidence that unphotolyzed ICN is reacting with some of the radical species generated photolytically. However, it is unclear why this process would be specific to experiments in DCM, and a more likely explanation is that a band associated with a shortlived intermediate overlaps the ICN bleach feature. One expected product of reaction of CN radicals with the solvent is HCN, which contributes an absorption band centered at approximately 2090 cm−1 that grows with time. This band is assigned to the fundamental vibrational transition from v1 = 0 → v1 = 1 (110) in HCN, with v1 denoting the CN stretching vibrational mode. The strong transient absorption feature centered at approximately 2060 cm−1 is assigned to the INC isomer of ICN,41,42,50,68 which forms by geminate recombination. Although the band intensity suggests a major product, the transition dipole moment of the CN stretching mode of INC is known to be much greater than for ICN in cryogenic matrices;50,68 as such, the intensity observed for the INC 110 transition is consistent with geminate recombination being a minor pathway for CN removal. Differences in transition dipole moments of this type make it difficult to draw more quantitative conclusions about the yields of products based on the relative intensities of bleach and absorption features in the TVAS spectra. The TVAS data for ICN photolysis in CH3CN and CD3CN show some clear similarities to the spectra for the chlorinated solvents as well as some differences. Two weak, sharp bands are evident in the TVAS data for ICN/CH3CN solutions at 2017 and 2043 cm−1 that are hard to discern from other spectral features for the chlorinated solvents. The 2043 cm−1 band also features in the ICN/CD3CN spectra, and the 2017 cm−1 band may be present but masked by a stronger 2010 cm−1 feature. The positions of these two bands, and their time dependence point to an assignment to vibrationally hot INC formed by the geminate recombination of I atoms with CN radicals. The INC v1 = 1 ← v1 = 0 absorption of the CN stretching mode occurs at 2069 cm−1; DFT calculations using the B3LYP functional with a 6-311G** basis set give an anharmonicity constant for the CN stretch of INC of x11 = −11.6 cm−1. The v1 = 2 ← v1 = 1 and v1 = 3 ← v1 = 2 vibrational hot bands are observed in acetonitrile at 2043 and 2017 cm−1, indicating an observed anharmonicity constant of x11 = −13 cm−1 that is consistent with the calculated value. The INC bands shift by approximately 5 cm−1 from lower to higher wavenumber in the first 5 ps, which may be a result of nascent excitation in the bending and N−I stretching modes of INC in addition to the CN stretch. The calculated anharmonicity constants coupling the bend (x12 = −3.4 cm−1) and I−N stretch (x13 = −1.6 cm−1) to the CN stretching mode of INC support this interpretation. A kinetic model accounting for the production and loss of the vibrationally excited CN stretch in INC is presented in section 3.4. TVA spectra of UV-excited ICN/ CHCl3 and ICN/CH2Cl2 solutions show similar signatures of vibrationally excited INC to the low wavenumber side of the

Table 1. Time Constants (in ps) for the Growth and Decay of Spectral Bands Observed by TEAS for UV-Excited ICN in Various Solvents time constant/psa free CN decay

340 nm feature growth

acetonitrile

0.6 ± 0.1

0.6 ± 0.1

chloroform

3.3 ± 1.0

1.7 ± 0.7

9.8 ± 1.5, 158 ± 25 1500 ± 150

dichloromethane

1.1 ± 0.5

1.1 ± 0.5

230 ± 50

solvent

ethanol methanol

340 nm feature decay

27 ± 3 23 ± 2

I*(2P1/2)− solvent CT band decay

2.0 ± 220 1.4 ± 190 1.5 ± 1.5 ±

0.4, ± 20 0.5, ± 20 0.8 0.8

a The band assignments are shown in Figure 2, and spectral decomposition and time-constant fitting used the KOALA program. Missing values indicate that no band associated with that particular feature was clearly observed. Where two time constants are reported, a biexponential decay function was required to fit the changing band intensity with time.

studied. Our time resolution prevents us from observing some of the faster processes reported previously by Rivera et al.39 The TEAS data report on the time scales for removal of CN radicals, but several competing pathways can account for these losses. These pathways are distinguished by the analysis of TVAS data presented in the next section. 3.3. Transient Vibrational Absorption Spectroscopy of Photoexcited ICN Solutions. TVAS of photoexcited ICN in various solvents using broadband mid-IR probe pulses concentrated on the CN stretching region around 1900− 2200 cm−1. Example TVAS data are presented in Figure 4 for measurements in acetonitrile (Figure 4a), acetonitrile-d (Figure 4c), DCM (Figure 4d), chloroform (Figure 4e), and THF (Figure 4f). The assignments of bands in spectra obtained in chlorinated solvents have been considered previously,42 and provide guidance for interpretation of spectra in other solvents. Figure 4b shows an example decomposition obtained after ICN photolysis in acetonitrile, with the analysis performed using the KOALA program.61 Our rationale for band assignments is explained below, and Table 2 provides a summary. The band centered at 2160 cm−1 in chloroform, and at 2169 cm−1 in DCM, is assigned to ICN on the basis of FTIR E

DOI: 10.1021/acs.jpca.5b10716 J. Phys. Chem. A XXXX, XXX, XXX−XXX

Article

The Journal of Physical Chemistry A

Figure 4. TVAS spectra obtained following ICN photolysis in (a) acetonitrile, (c) acetonitrile-d, (d) dichloromethane, (e) chloroform, and (f) tetrahydrofuran. Panel b illustrates decomposition of a single time spectrum in acetonitrile and assignment of the constituent absorption features. Each spectral feature is fitted with a skewed Gaussian function which matches the shape of the late-time INC 110 absorption band unless stated otherwise in the text. Spectra corresponding to individual time delays are represented with different colored spectra from purple to dark red and show the evolution from reactants to products up to maximum time delays of 3000 ps.

Table 2. Band Center Wavenumbers Observed in TVA Spectra Obtained in the CN Stretching Region Following ICN Photolysis in Five Solvents absorption band center/cm‑1 a solvent acetonitrile acetonitrile-d chloroform dichloromethane tetrahydrofuran a

ICN

110

2166 2166 2160 2169

solvent

HCN

110

HCN 121

HCN 132

INC 110

INC 121

INC 132

radical adduct

2043 2043 2049 2039

2018 2018 2023 2013

2036 2012

2039

2069 2069 2075 2065 2061

2094 2120 2098 2096 2080

2060

The spectral resolution of 2.6 cm−1 determines the uncertainties in the specified wavenumber values.

acetonitrile is discussed in greater detail in a companion paper.69 The HCN 110 band is not observed in CD3CN, as expected in the absence of any hydrogenated species, but is only weakly evident in the CH3CN solutions. The weakness of the HCN 110 absorption relative to the ICN bleach in acetonitrile solutions, when compared to observations for other solvents, suggests that CN radicals do not react efficiently with acetonitrile to make HCN on these short time scales. However, the development of the 2120 cm−1 bleach in the CD3CN spectra demonstrates that some reaction with the solvent is occurring. These observations can be reconciled if there is an alternative reaction pathway that competes with H/D atom abstraction. The bands at 2036 cm−1 in CH3CN and 2012 cm−1 in CD3CN, labeled as radical adducts in Figure 4, parts a and c, support this interpretation because they have a delayed onset of ∼10 ps commensurate with the time scale for the loss of the CN radicals observed by TEAS. The delay suggests that an intermediate to the formation of the species responsible for the 2036 and 2012 cm−1 absorptions in CH3CN and CD3CN is

INC fundamental band, although such an assignment does not fully account for the absorptions in this region. The CN stretch excited hot bands of INC in the chlorinated solvents are separated by the same anharmonicity constant of x11 = −13 cm−1 observed in acetonitrile, although the peak positions show different solvent-induced shifts. An additional bleach of a solvent vibrational band is observed in CD3CN at 2120 cm−1. The gradual increase in the depth of this feature is attributed to loss of solvent molecules by reaction with CN radicals. The growth in intensity of the bleach can be fitted with a single exponential time constant of 12 ± 3 ps which is more than an order of magnitude slower than the decay of free CN observed in our TEAS studies. The 12 ± 3 ps growth in the CD3CN bleach therefore demonstrates that CN reactions with the solvent must have an intermediate step, which we suggest corresponds to formation of CN−solvent complexes. These complexes react more slowly than the time scale for free CN decay, and account for the observed 12 ps CD3CN removal. The short-time chemistry of CN radicals in F

DOI: 10.1021/acs.jpca.5b10716 J. Phys. Chem. A XXXX, XXX, XXX−XXX

Article

The Journal of Physical Chemistry A a direct result of reaction between the solvent and CN radicals. The spectroscopically observed species subsequently grow with ∼30 ps time constants and persist out to the longest time scales measured. Ab initio calculations of vibrational frequencies of possible reaction products indicate that the 2036 cm−1 band derives from CN-addition to the N or C atom of the nitrile group in acetonitrile to produce a C3H3N2• radical. Possible structures for the CN−acetonitrile radical adduct are discussed elsewhere.69 The TVA spectra obtained for ICN/THF solutions are strikingly different to those obtained with other solvents. For example, the INC band is weak compared to the HCN 110 band, the intensity of which is amplified by an enhanced transition dipole moment of HCN in THF (see Supporting Information). Furthermore, the INC absorption is overlapped by an absorption centered at 2060 cm −1 that we assign to vibrationally hot HCN. This overlapping band grows and decays before the INC 110 band fully develops, producing a band that peaks at 2061 cm−1. An additional small absorption feature is observed at 2039 cm−1, which is also assigned to internally excited HCN. The HCN products of reactions of CN radicals with organic molecules in solution are known to be excited in the C−H stretching (v3) and bending (v2) modes41,58 from TVAS observations in the C−H stretching region around 3200 cm−1. In the gas phase, the couplings of these modes to the CN stretch are quantified by anharmonicity constants x12 = −3.15 cm−1 and x13 = −14.9 cm−1.70 While hot HCN bands in the nitrile region could arise from the 110 absorption band in otherwise C−H stretch and bend excited HCN, the 2060 and 2039 cm−1 band positions observed in the TVA spectra following ICN photolysis in THF agree better with expectations for the 121 and 132 hot bands of CN stretch excited HCN, for which the anharmonicity constant is x11 = −10.3 cm−1. This assignment is therefore made in Table 2, and we return to it and the analysis of the time dependences of the bands in section 3.5. An additional band is observed at 2048 cm−1 in THF solution; this is not reproducibly present in every data set and its assignment therefore remains uncertain. 3.4. Vibrational Cooling of INC. The TVA spectra reported in section 3.3 provide evidence that INC is produced vibrationally hot by geminate recombination of the CN and I radicals. The anharmonic shifts predicted by electronic structure calculations point toward excitation of the CN stretching mode and we assign the observed bands accordingly, although the other modes are also likely to be excited by formation of the I−N bond. The INC(v1) vibrational cooling rates deduced from the decays of the v1 = 2 ← v1 = 1 and v1 = 3 ← v1 = 2 hot bands and growth of the fundamental v1 = 1 ← v1 = 0 absorption band depend strongly on the solvent. The cooling kinetics were treated using a stepwise vibrational cooling model in which only single vibrational quantum changes were considered. The Landau−Teller model of vibrational relaxation predicts that the time constants τn→(n−1) for the v = n→(n − 1) steps down a ladder of vibrational levels should scale according to τn→(n−1) = (τ1→0/n).71 This expectation was used as a constraint in kinetic fits that used the model shown in Scheme 1 to describe the INC vibrational cooling. The model assumes that INC can be produced in any of the probed vibrational levels up to v1 = 2 directly by the recombination reaction. Fits were performed to time-dependent band intensities, which depend on both the population differences between the two vibrational levels connected by an IR transition, and on the increase in transition

Scheme 1

Figure 5. Fits (solid lines) to kinetic Scheme 1 of the TVAS band intensities (dots) for INC formed through geminate recombination of I and CN for (a) acetonitrile, (b) dichloromethane, and (c) chloroform solutions. The fitting was performed simultaneously for three bands of INC: red, 110; blue, 121; green, 132.

moments as the vibrational state probed increases. Figure 5 shows the fits to this model for band intensities measured in acetonitrile (Figure 5a), dichloromethane (Figure 5b) and chloroform (Figure 5c) solutions, with the derived time constants for vibrational relaxation presented in Table 3. The relative magnitudes of the time constants for the association steps derived from the fits account for the relative band intensities and depend inversely on the nascent branching between INC CN stretching vibrational levels. The recombination time scales derived from the fits presented in Figure 5 for different INC vibrational levels are summarized in the Supporting Information. The combined time constant of 10 ps for recombination in acetonitrile (obtained using 1/τ = Σi(1/τi)) is consistent with fast decay G

DOI: 10.1021/acs.jpca.5b10716 J. Phys. Chem. A XXXX, XXX, XXX−XXX

Article

The Journal of Physical Chemistry A

Table 3. Branching Ratios for the Geminate Recombination to Particular Vibrational Levels of INC and Time Constants for Vibrational Relaxation Steps, Obtained from Fits of TVA Spectral Data to the Kinetic Model of Scheme 1 recombination branching ratio

relaxation time constant/ps

solvent

INC (v1 = 2)

INC (v1 = 1)

INC (v1 = 0)

τ2→1

τ1→0

acetonitrile dichloromethane chloroform

0.08 ± 0.04 0.12 ± 0.05 0.12 ± 0.03

0.26 ± 0.04 0.32 ± 0.08 0.32 ± 0.05

0.66 ± 0.08 0.56 ± 0.13 0.56 ± 0.10

150 ± 23 330 ± 75 700 ± 120

300 ± 45 660 ± 150 1400 ± 250

separate publication.69 Here, we focus on the HCN-forming channel. 3.5.1. Reaction of CN Radicals with Dichloromethane. The spectra obtained in the CN stretching region following ICN photolysis in DCM solution, shown in Figure 4d, were supplemented by spectra of the type we have reported previously in the C−H stretching (ν3) region around 3260 cm−1 for other reactions.41,42 In this region, C−H stretching excitation shifts the IR absorption bands of HCN by close to −100 cm−1 relative to the fundamental band for each additional quantum of vibration. HCN that is vibrationally excited in the C−H mode can therefore be clearly distinguished from ground vibrational level molecules in the TVAS data. TVAS spectra obtained in the C−H stretching region following 267 nm photolysis of ICN in dichloromethane were analyzed to obtain the time-dependent band intensities shown in Figure 6. These

Table 4. Single and Double Exponential Time Constants for Growth in Intensity of Product Bands Corresponding to HCN and the Addition Products Observed at 2036 cm−1 in Acetonitrile and 2012 cm−1 in d-Acetonitrile time constant/ps solvent

HCN formation

radical adduct formationa

acetonitrile acetonitrile-d chloroform dichloromethane tetrahydrofuran

73 ± 17

7 ± 3, 33 ± 4 10.9 ± 3.5, 32 ± 5

700 ± 90 900 ± 120 15 ± 2

a

Two time constants are quoted for the growth of the radical adduct; the first accounts for the delayed onset of the radical adduct species.

time constants obtained from the TEA spectra for the 340 nm feature after BrCN (11 ± 2 ps) and ICN (9.8 ± 1.5 ps) photolysis in acetonitrile. However, given the limited ICN bleach recovery observed in each solvent presented in Figure 4 and the large transition moment of the v1 band of INC, recombination appears to be a minor pathway. The recombination time scales are therefore limited by the faster loss of CN radicals through other channels. Absolute recombination yields are not deduced from the fits to the kinetic model, but the branching ratios to different INC vibrational levels depend inversely on the fitted recombination time constants and are presented in Table 3. These branching ratios appear to be insensitive to the choice of solvent for experiments in chloroform, dichloromethane and acetonitrile. Excitation of the CN stretching mode is perhaps surprising given that the new bond formed by geminate recombination is the I−N bond, whereas the CN bond might be regarded as a spectator. Nonetheless, multiple quanta of excitation of the CN stretch are observed in all of the solvents investigated although the recombination favors v1 = 0 (Table 3). The UV photolysis of ICN produces